pubmed.ncbi.nlm.nih.gov

NAMPT overexpression induces cancer stemness and defines a novel tumor signature for glioma prognosis - PubMed

  • ️Sun Jan 01 2017

NAMPT overexpression induces cancer stemness and defines a novel tumor signature for glioma prognosis

Antonio Lucena-Cacace et al. Oncotarget. 2017.

Abstract

Gliomas are the most prevalent primary malignant brain tumors associated with poor prognosis. NAMPT, a rate-limiting enzyme that boosts the nicotinamide adenine dinucleotide (NAD) regeneration in the salvage pathway, is commonly expressed in these tumors. NAD metabolism is required to maintain tissue homeostasis. To maintain metabolism, cancer cells require a stable NAD regeneration circuit. However, high levels of NAD confer resistance to therapy to these tumors, usually treated with Temozolomide (TMZ). We report that NAMPT overexpression in glioma cell lines increases tumorigenic properties controlling stem cell pathways and enriching the cancer-initiating cell (CIC) population. Furthermore, NAMPT expression correlated with high levels of Nanog, CD133 and CIC-like cells in glioblastoma directly extracted from patients. Meta-analysis reveals that NAMPT is also a key factor inducing cancer stem pathways in glioma cells. Furthermore, we report a novel NAMPT-driven signature which stratify prognosis within tumor staging. NAMPT signature also correlates directly with EGFR positive and IDH negative tumors. Finally, NAMPT inhibition increases sensitivity to apoptosis in both NAMPT-expressing cells and tumorspheres. Therefore, NAMPT represents a novel therapeutic target in Glioma progression and relapse.

Keywords: NAMPT; cancer initiating cell; gene signature; glioblastoma; glioma.

PubMed Disclaimer

Conflict of interest statement

CONFLICTS OF INTEREST Authors declare no conflicts of interest.

Figures

Figure 1
Figure 1. NAMPT expression correlates with tumor grade

(A) Analysis of Normal Brain [n = 216], Glioma [n = 577], Pilocytic Astrocytoma [n = 41] and N euroblastoma [n = 45] datasets available through the R2 database indicates that NAMPT is significantly upregulated in tumors derived from the ectoderm, such as pilocytic astrocytoma [*p = 0.01 with ANOVA compared to normal brain] and glioma [*p = 0.01 with ANOVA compared to normal brain]. (B) Analysis of normal brain [n = 23] and glioblastoma [n = 81] datasets from the Sun database available in Oncomine indicates that NAMPT is highly overexpressed in glioblastoma [**p < 0.01 with ANOVA compared to normal brain]. (C) New retrospective analyses of new databases, comparing normal brain GSE13564, GSE11882 [n = 216], Glioblastoma [GSE4290, GSE16011 Glioblastoma series taken together; n = 159], Anaplastic Astrocytoma - glioma grade III [GSE4290, Astrocytoma grade III GSM series; n = 16] and Glioblastoma [n = 9; GSM379855, GSM379856, GSM379857, GSM379858, GSM379870, GSM379871, GSM379872, GSM379873 and GSM379874] cell lines shows that NAMPT is upregulated in the mRNA pools of these cell lines [GSE15209 ****p < 0.0001 with ANOVA compared to normal brain] and that its expression is upregulated at baseline in glioma grade III cells [***p < 0.001 with ANOVA compared to normal brain]. (D–E) The Oncomine glioma dataset shows a correlation between NAMPT expression and tumor grade. (D) Analysis of the Oncomine Glioma dataset stratifying NAMPT according to Glioma grade with Grade II [n = 45] Grade III [n = 31] and grade IV [n = 81] tumors shows a significant difference in NAMPT expression between grade II and grade IV tumors [**p < 0.001]. (E) A second database analyzed in Oncomine strengthens the data previously shown: NAMPT expression correlating with tumor grade: Grade II [n = 50], Grade III [n = 26], Grade IV [n = 77] shows a significant difference in NAMPT expression between grade II and grade IV tumors [**p = 0.0052] (F) Analysis of TCGA Glioblastoma molecular subtypes Proneural, Neural, Classical and Mesenchymal.

Figure 2
Figure 2. NAMPT is an independent indicator of glioma patient outcomes

Analysis of glioma datasets available through Oncomine and R2 indicates the existence of a significant correlation between high NAMPT expression and poor survival in the French tumor glioma dataset (A) [n = 124 NAMPT low; n = 149 NAMPT high; p = 3,20E- 13 with log-rank analysis]; Kawaguchi tumor glioma dataset. (B) [n = 31 NAMPT low; n = 19 NAMPT high; p = 1.3E-05 with log-rank analysis]; Paugh tumor glioma dataset. (C) [n = 13 NAMPT low; n = 34 NAMPT high; p = 0.054 with log-rank analysis]; French tumor glioma dataset, subtype grade II. (D) [n = 9 NAMPT low; n = 15 NAMPT high; p = 0.172 with log-rank analysis]; French tumor glioma dataset, subtype grade III. (E) [n = 40 NAMPT low; n = 45 NAMPT high; p = 0.000018 with log-rank analysis]; and French tumor glioma dataset, subtype grade IV. (F) [n = 85 NAMPT low; n = 71 NAMPT high; p = 0.000360 with log-rank analysis]. The poor prognosis of grade IV glioblastoma is confirmed via analysis of the TCGA 540 glioblastoma dataset [n = 96 NAMPT low; n = 408 NAMPT high; p = 0.000071 with log-rank analysis] (G), TCGA 395 glioblastoma dataset [n = 68 NAMPT low; n = 309 NAMPT high; p = 0.00089 with log-rank analysis] (H) and French Exon-Core dataset [n = 85 NAMPT low; n = 10 NAMPT high; p = 0.0026 with log-rank analysis] (I).

Figure 3
Figure 3. NAMPT expression increases tumorigenic and CIC properties

(A, B) Q-RT-PCR shows NAMPT overexpression in either SF268 (A) or U251MG (B). (C, F) Western blot analysis shows NAMPT overexpression and NAMPT silencing with shRNA in SF268 (C) and U251MG (D). (E, F) Analysis of the growth curve indicates that NAMPT overexpression confers a proliferative advantage [*p < 0.05; **p < 0.01 with ANOVA compared to vector], whereas NAMPT underexpression slows proliferation [*p < 0.05, **p < 0.01 with ANOVA compared to vector]. (G, H) Clonogenicity assay results indicate that NAMPT overexpression increases the number of colonies [**p < 0.01 with ANOVA compared to vector], whereas NAMPT underexpression decreases the number of colonies [*p < 0.05 with ANOVA compared to vector]. (I) Tumorsphere-forming assay in both cell lines indicates that NAMPT overexpression increases both number and size, whereas NAMPT underexpression decreases tumorsphere number and size. (J, K) Analysis of clone phenotypes [holoclones – black, meroclones – gray, paraclones – white] shows that NAMPT overexpression increases the number of holoclones [**p < 0.01 with ANOVA compared to vector], whereas NAMPT underexpression decreases the holoclone number [*p < 0.05 with ANOVA compared to vector]. J shows data from SF268 cell line. K shows data from U251MG cell line. (L) CD133 analysis with FACS indicates that NAMPT overexpression increases CD133 levels [*p < 0.05 with ANOVA compared to vector], whereas NAMPT underexpression decreases CD133 levels [**p < 0.01 with ANOVA compared to vector]. (M) CD44 analysis with FACS indicates that NAMPT overexpression increases CD44 levels [**p < 0.01 with ANOVA compared to vector], whereas NAMPT underexpression decreases CD44 levels [*p < 0.05 with ANOVA compared to vector]. (N–Q) Single-cell Sphere-forming assay indicates that NAMPT overexpression increases both number and size, whereas NAMPT underexpression decreases tumorsphere number and size. (N) Single cell tumorsphere forming efficiency representative picture. (O) Single cell tumorsphere forming efficiency percentage. (Q) Single cell tumorsphere size. (P) Single cell colony (full culture) forming efficiency percentage.

Figure 4
Figure 4. NAMPT expression correlated with high levels of cancer stem cell-like cells in glioblastoma extracted directly from patients

(A) We first took 14 glioblastoma tumor samples directly from patients as well as matched non-tumor samples and generated a tissue microarray, TMA. These TMAs were stained for NAMPT and Nanog according to M&M. In these samples we evaluated the expression of NAMPT and Nanog, as surrogated marker of cancer stem cell-like levels of these tumors. (B) Levels of NAMPT and (C) Nanog were related in matched samples, from the same patient. (D) Evaluation of the correlation between matched samplesof NAMPT and Nanog. There was a clear correlation (pearson r = 0.53, p < 0.01) between the expression of Nanog and NAMPT. (E) We took 5 fresh glioblastoma samples from patients. After tissue disaggregation, we directly measured NAMPT levels by Q-RT-PCR and parallely seeded 3000 cells to measure the number of tumorspheres formed. Then we plotted to establish 1 to 1 correlation, between the level of NAMPT and the number of tumorspheres. We found a strong direct correlation of the tumorsphere number and NAMPT levels for each tumor.

Figure 5
Figure 5. NAMPT activates EMT and stem pathway effectors

(A) Q-RT-PCR analysis of NAMPT overexpression in empty vector [**p < 0.01 with t-test]. Q-RT-PCR analysis of stem cell gene expression in empty vector: NANOG [***p < 0.001 with t-test], SOX2 [*p < 0.05; **p < 0.01 with t-test], OCT4 [**p < 0.01; ***, p < 0.001 with t-test] and CD133 [*p < 0.05; **p < 0.01 with t-test]. (B) RT-qPCR analysis of EMT gene expression in empty vector: FOXC2 [**p < 0.01 with t-test], TWIST1 [*p < 0.05; **p <0.01 with t-test], VIM [*p <0.05; **p <0.01 with t-test] and SNAI1 [**p < 0.01 with t-test]. (C) Analysis of the Yamanaka dataset (GSM241846) with respect to induced pluripotent stem cell (iPSC) generation from human dermal fibroblasts (hDF) demonstrates an increase in NAMPT levels with reprogramming [*p = 0.01; **p < 0.01 with ANOVA compared to human dermal fibroblasts]. (D) Analysis of the Yamanaka dataset (GSE15148) with respect to iPSC generation from mouse embryonic fibroblasts (mEF) demonstrates a significant increase in NAMPT expression with reprogramming. Mouse embryonic stem cells (mESC) also express high levels of NAMPT compared to fibroblasts [*p = 0.01; **p < 0.01 with ANOVA compared to mouse embryonic fibroblasts]. (E) Analysis of the Thomson dataset (GSE5259) with respect to iPSC generation from human foreskin fibroblasts (parent) demonstrates an increase in NAMPT levels with reprogramming. Human embryonic stem cell lines (H1L and H7) also express high levels of NAMPT compared to fibroblasts [*p = 0.01; **p < 0.01 with ANOVA compared to human foreskin fibroblasts]. V: cells expressing vector only. NAMPT: Cells overexpressing ectopic NAMPT cDNA.

Figure 6
Figure 6. NAMPT induces a gene signature that correlates with glioma tumor grade and predicts poor survival

(A) R2 expression analysis of the gene signature triggered by NAMPT in the Sun dataset shows a correlation between NAMPT expression and advanced glioma stage: grade IV. R2 expression analysis of the gene signature triggered by NAMPT showed genes related to Stem Cell phenotype. (B) pathways controlling CIC-like propersies related to NAMPT expression. (C) Genes associated to NAMP gene expression [all p < 0.01, Pearson r is shown in each case], and the pathway each gene is associated to. (D) Overall survival probability comparing the patients showing low expression signature vs high expression signature. R2 expression analysis of the gene signature triggered by NAMPT in the Sun Brain tumor database in a biased cohort of patients with high signature expression [n = 10] and low signature expression [n = 10] shows poor survival in patients [p = 0.06 with log-rank analysis]. (E) Analisis of the expression of the different genes of the signature in SF268 and U2521MG expressing NAMPT.

Figure 7
Figure 7. Analysis of NAMPT-related signature in the TCGA database

(A) Nampt-derived signature is represented. (B). Relative expression leves of each gene are represented in normal brain vs glioma samples. (C) Nampt-derived signature levels clasiffy Grade IV gliobastoma samples. (D) Nampt-derived signature levels classify the different subtypes of Glioblastoma according to the whole dataset of Glioblastoma TCGA dataset. (E–F) Overall survival probability comparing the patients showing low expression signature vs high expression signature. R2 expression analysis of the gene signature triggered by NAMPT in the TCGA 540 (E) or French (F) glioblastomma databases in a biased cohort of patients with high signature expression [n = 10] and low signature expression [n = 10] shows poor survival in patients [p < 0.01 with log-rank analysis].

Figure 8
Figure 8. NAMPT induces a gene signature that correlates with WT IDH1 patients and EGFR positive tumors

To this study we have selected the French database that contains information about the IDH1 and EGFR mutational status. (A) Nampt-derived signature is present in a high percentage of Grade IV gliobastoma samples from French glioma database. (B) Nampt-derived signature predicts an enrichment of IDH1 mutatins in patients negative for our signature, indicating a better prognosis in patients with IDH1 mutations. (C) On the other hand, most, if not all, gliomas with EGFR amplifications strong positive correlation with our NAMPT-derived signature.

Similar articles

Cited by

References

    1. Zeng T, Cui D, Gao L. Glioma: an overview of current classifications, characteristics, molecular biology and target therapies. Front Biosci (Landmark Ed) 2015;20:1104–1115. - PubMed
    1. Le Rhun E, Taillibert S, Chamberlain MC. Anaplastic glioma: current treatment and management. Expert Rev Neurother. 2015;15:601–620. - PubMed
    1. Reardon DA, Wen PY. Glioma in 2014: unravelling tumour heterogeneity-implications for therapy. Nat Rev Clin Oncol. 2015;12:69–70. - PubMed
    1. Morokoff A, Ng W, Gogos A, Kaye AH. Molecular subtypes, stem cells and heterogeneity: Implications for personalised therapy in glioma. J Clin Neurosci. 2015;22:1219–1226. - PubMed
    1. Sorensen MD, Fosmark S, Hellwege S, Beier D, Kristensen BW, Beier CP. Chemoresistance and chemotherapy targeting stem-like cells in malignant glioma. Adv Exp Med Biol. 2015;853:111–138. - PubMed